Prediction of mortality in community-living frail elderly people with long-term care needs

OBJECTIVES: To develop and validate a prognostic index for mortality in community-living, frail elderly people. DESIGN: Cohort study of Program of All-Inclusive Care for the Elderly (PACE) participants enrolled between 1988 and 1996. SETTING: Eleven PACE sites, a community-based long-term care program that cares for frail, chronically ill elderly people who meet criteria for nursing home placement. PARTICIPANTS: Three thousand eight hundred ninety-nine PACE enrollees. The index was developed in 2,232 participants and validated in 1,667. MEASUREMENTS: Time to death was predicted using risk factors obtained from a geriatric assessment performed by the PACE interdisciplinary team at the time of enrollment. Risk factors included demographic characteristics, comorbid conditions, and functional status. RESULTS: The development cohort had a mean age of 79 (68% female, 40% white). The validation cohort had a mean age of 79 (76% female, 65% white). In the development cohort, eight independent risk factors of mortality were identified and weighted, using Cox regression, to create a risk score: male sex, 2 points; age (75-79, 2 points; 80-84, 2 points; >= 85, 3 points); dependence in toileting, 1 point; dependence in dressing (partial dependence, 1 point; full dependence, 3 points); malignant neoplasm, 2 points; congestive heart failure, 3 points; chronic obstructive pulmonary disease, 1 point; and renal insufficiency, 3 points. In the development cohort, respective 1- and 3-year mortality rates were 6% and 21% in the lowest-risk group (0-3 points), 12% and 36% in the middle-risk group (4-5 points), and 21% and 54% in the highest-risk group (> 5 points). In the validation cohort, respective 1- and 3-year mortality rates were 7% and 18% in the lowest-risk group, 11% and 36% in the middle-risk group, and 22% and 55% in the highest-risk group. The area under the receiver operating characteristic curve for the point score was 0.66 and 0.69 in the development and validation cohorts, respectively. CONCLUSION: A multidimensional prognostic index was developed and validated using age, sex, functional status, and comorbidities that effectively stratifies frail, community-living elderly people into groups at varying risk of mortality.